GIS Lecture 11: Spatial Analyst

1. GIS Lecture 11: Spatial Analyst

2. Outline • Spatial Analyst Overview • Grid Maps • Raster Layers • Raster Masks • Kernel Density Smoothing • Site Suitability Study • Poverty Risk Model

3. Spatial Analyst Overview

4. Spatial Analyst • Traditional GIS • Spatial queries, buffer analysis, overlay analysis • Spatial Analyst • Quantify spatial patterns and relationships • Track changes over time and determine patterns • Identify anomalous spatial patterns using mathematical statistics • Statistical significance and verifying conclusions

5. Primary Features • Data analysis • Allows rapid generation of maps that are based on complex statistics • Surface modeling • Creates predicted surface (grid maps) from unmeasured points based on statistical analysis of measured points

6. Load Spatial Analyst Extension • Tools, Extensions…

7. Grid Maps

8. Grid Maps • Divides geographic space into uniform blocks called cells • Every cell represents a certain specified portion of the earth, such as a square mile, kilometer, meter, etc. • Each cell is given a value that describes the site, such as elevation, land use type, number of crimes, etc.

9. Grid Map Example • Housing units in City of Pittsburgh • Census Block Centroids • Kernel Density Raster Map

10. Raster Layers

11. Raster Basemap • Free from U.S. Geological Survey (USGS) • Digital Elevation Model (DEM) • NED shaded relief, 1/3 arc second

12. Raster Basemap • Land use map • NLCD 2001

13. Raster Layer Properties • All raster maps are rectangular coordinate system

14. Create Raster Mask • Conversion Tools • Feature to Raster

15. Create Grid Map • Convert TIF to Grid • Export TIF

16. Environment Settings • Settings for environment tools • Set once to use throughout project

17. Extract Raster Masks • Extract (Clip) Land Use to Pgh Mask

18. Hillshade Raster • Hillshade function simulates illumination of a surface • Gives 3D appearance • Parameters • Altitude of light source above surface’s horizon • Angle (azimuth) relative to true north

19. Create Hillshade Raster • DEM to Hillshade

20. Resultant Hillshade Layer • Original hillshade layer Modified hillshade layer

21. Kernel Density Smoothing

22. Kernel Density Smoothing • Analyzes out of heart attack incidences in Pittsburgh • Uses point shapefile of census block centroids and point shapefile of out of hospital cardiac arrests (OHCA shapefile) • Estimates the incidence of heart attacks per unit area (density) • Two parameters • Cell size • Search radius

23. Existing Maps • OHCA and existing raster land use map • Heart attack locations are in developed areas

24. Existing Maps • Point centroids with population • OHCA points

25. Density Map • Create density map for heart attack incidence • Pittsburgh blocks average 300 ft per side in length • 2.5 blocks reasonable for defibrillator locations • Look at areas 5 blocks by 5 blocks or 1,500 ft • 150 ft cell size

26. Resultant Density Map • Shown with standard deviation and custom color ramp

27. Resultant Density Map • OHCA added • Note cluster in Downtown Pittsburgh where people work but don’t live

28. Raster Value Points • Extracts point estimates from raster surface for each OHCA point • Use extracted densities multiplied by block areas to estimate number of heart attacks • Better predictor of heart attacks

29. Extract Raster Value Points • Extraction Toolset

30. Resultant Layer • OHCAPredicted shapefile • Has attribute value, RASTERVALU, which is an estimate of heart attack density per sq foot, in the vicinity of each block

31. Calculate Predicted Heart Attacks • Resultant estimate will be larger than actual number of heart attacks in OCHA (these are only ones with bystander help) • Expression to calculate prediction • 5 * [RASTERVALU] * [AREA] • “five” is for 5 year sample of heart attack data in OHCA

32. Scatter Plot • Actual vs. predicted heart attacks • Pittsburgh has 7,466 blocks • 1,509 blocks with heart attacks

33. Resultant Graph • At the scale of blocks, predicted values correlate poorly with actual values • Factors other than population residence is needed

34. Site Suitability Study

35. Site Suitability Study • Evaluates other factors for heart attack predictability (e.g. commercial land use) • Uses 600 foot buffer around commercial areas

36. Convert Buffers to Raster

37. Kernel Density Map • Create a kernel density map to use for calculations

38. Calculate Simple Query • Query kernel density map (HeartAttack) for areas that have high density and merit a defibrillator • 25 block area(1,500 ft x 1,500 ft , 2.25 x 106 sq ft of area) • 10 or more heart attacks every 5 years in locations where bystander help is possible • Heart attack density : • 10 heart attacks/ 2.25 x 106 sq ft =/0.000004444

39. Calculate Simple Query • Raster Calculator

40. Simple Query Result

41. Calculate Compound Query • Add second criteria: commercial buffer

42. Compound Query Result

43. Poverty Risk Model

44. Model for Risk Index • Poverty study as another risk for heart attacks • Population below poverty income line • Female headed household with children • Population with less than HS education • Workforce who are unemployed • Improper linear model for poverty • Calculate Z-score values • Data selected by subtracting the mean and dividing by the standard deviation for above variables, then averaging them

45. Risk Model Base Layers • Block Groups • NoHighSch2 (ho high school degree) • Male16Unem (males in workforce who are unemployed) • Poverty (population below poverty income) • Blocks • FHH (female headed households with children)

46. Risk Model Base Layers • Difficult to represent using vectors

47. Create New Model • New toolbox and model

48. Kernel Density Layers • Create kernel density layer for first input

49. Kernel Density Layers • Create kernel density layer for additional input

50. Standard Deviation Statistics • Mean and standard deviation for raster layers